Kidins220 correlates with tau in Alzheimer’s disease brain and cerebrospinal fluid

Identification of neurodegeneration-monitoring biomarkers would be of great clinical value for Alzheimer´s disease (AD) diagnosis. Using N- or C-terminal antibodies, we studied the pro-survival synaptic effector, Kidins220, in the brain and cerebrospinal fluid (CSF) of controls and AD patients. Only the N-terminal antibody showed a positive correlation between Kidins220 and phosphorylated-tau in AD brains. Using this antibody, Kidins220 was detected in CSF from AD patients where it positively correlated with CSF phosphorylated-tau and tau. This study highlights the potential of Kidins220 to be used as a CSF biomarker in AD.


INTRODUCTION 1
Alzheimer's disease (AD), the most prevalent age-related dementia, is characterized by progressive neurodegeneration and severe synaptic and neuronal loss.
Neuropathological hallmarks of AD are extracellular senile plaques containing amyloid-β (Aβ) and intracellular neurofibrillary tangles enriched in hyperphosphorylated-tau [1,2]. Currently, reduced Aβ 1-42 and increased phosphorylated-tau (p-tau) and tau levels in the cerebrospinal fluid (CSF) are used as diagnostic markers for AD [3,4]. However, additional biomarkers for the early stages of AD pathogenesis could help improve AD diagnosis and monitor neurodegeneration. To this end, synaptic proteins in the CSF represent good candidates for further improving the prognostic accuracy of the AD biomarker panels [4].
Kinase D interacting substrate of 220 kDa (Kidins220) [5], also known as ankyrin repeat-rich membrane spanning "ARMS" [6], is an integral membrane protein present at the synapse where it is critical for neurotrophin and NMDARs signaling and neuronal survival [7,8]. Dysfunctional neurotrophic support, synaptopathy and excitotoxicity (pathological overstimulation of the glutamate N-methyl-D-aspartate receptors; NMDARs) are involved in AD neurodegeneration [9], making Kidins220 a promising candidate.
We have previously demonstrated an increase in Kidins220 in brains from AD patients concomitantly with Braak stage progression [10]. We also observed that, at  [10]. However, despite this initial observation, a thorough study of the correlation between Kidins220 and tau accumulation in the AD brain has yet to be reported.
Here we have evaluated Kidins220 levels in AD and control brain samples, examining the potential correlation with tau. Due to the potential for Kidins220 as a biomarker for underlying neurodegeneration, we have also analyzed Kidins220 levels in CSF samples from our cohort of patients.

Clinical cohort
Subjects from the Memory Unit (Hospital Santa Creu I Sant Pau) underwent formal cognitive evaluation [13] by neurologists with expertise in neurodegenerative diseases. Cognitively healthy control subjects showed results within the normal range and were negative for the core AD biomarkers based on our in-house criteria (CSF Aβ 1-42 >550ng/ml, CSF t-tau <350ng/ml or CSF p-tau <61ng/ml) [14]. Patients with amnesic mild cognitive impairment (aMCI) were diagnosed according to NIA-AA criteria [15]. Patients with aMCI who were also positive for AD biomarkers (CSF Aβ 1-42 <550ng/ml, CSF t-tau >350ng/ml or CSF p-tau >61ng/ml) were classified as Prodromal AD due to the increased probability of conversion to AD [15]. Patients who met the criteria of the National Institute of Neurological and Communicative Disorders and Stroke and the Alzheimer's Disease and Related Disorders Association [16] and were positive for AD biomarkers were classified as typical AD. All participants gave their written consent, and the study was approved by the local ethics committee following the ethical standards recommended by the Helsinki Declaration.

CSF collection and assessment
Collection of CSF was achieved by lumbar puncture according to international consensus recommendations [17]. Quantification of AD CSF biomarkers by ELISA was performed as previously described [14].

Quantification and Statistical analysis
Full-length (FL) Kidins220 was quantified by densitometric analysis (NIH Image) after immunoblotting with Kidins220-Ct. Quantification values using Kidins220-Nt resulted from the sum of signals from FL and N-terminal fragments in brain or only FL in CSF.

7
Kidins220 levels were normalized to NSE in brain samples and to Ponceau staining in CSF samples, and expressed relative to the mean of controls. Signal for p-tau was normalized to tau and to NSE in brain samples. Student's t-test was used to compare Kidins220 levels between controls and patients. Correlations between Kidins220, ptau, tau and Aβ 1-42 were determined by Spearman's Correlation test. Statistical analyses were performed using GraphPad Prism (San Diego, CA, USA).

Short post-mortem interval and detection of Kidins220 N-terminal region are critical factors to study Kidins220 in human brain.
Initial studies on Kidins220 in AD were carried out on post-mortem brain samples using an antibody raised against the last 17 aminoacids of Kidins220 C-terminus (Kidins220-Ct) [10]. We have previously shown that Kidins220 very C-terminal end is lost after cleavage by the protease calpain in models of excitotoxicity and cerebral ischemia, and that cleaved-forms of the protein could only be detected by an antibody recognizing the N-terminal region (Kidins220-Nt) [7,18]. Therefore, Kidins220-Ct signal could be severely diminished in situations where this protease is highly active, whilst the N-terminal signal may be preserved (see Figure 1A for details of antibodies and calpain-cleaved Kidins220 N-terminal fragments).
Because calpain is activated in post-mortem tissue [19], we first evaluated the influence of post-mortem interval (PMI) on the ability of C-terminal and N-terminal antibodies to identify Kidins220 in human brain tissue. In order to strictly study the effect of increasing PMI, control and AD brain samples of short PMI (6 and 5h, respectively) were left at room temperature for additional periods of time before analysing Kidins220 by immunoblot ( Figure 1B). Both antibodies detected a band corresponding to full-length (FL) Kidins220 but, as time increased, FL signal 8 disappeared and N-terminal fragments emerged, being still visible after 24h only with Kidins220-Nt ( Figure 1B). Kidins220-Ct did not detect any proteolytic band (not shown).
With this in mind, we examined Kidins220 levels using these two antibodies in brain necropsies over a range of PMI from AD patients and control individuals (see Supplementary Table 1  Kidins220 positively correlates with p-tau in human AD brain necropsies using a novel N-terminal antibody. Next, we compared Kidins220 levels between AD and control brain necropsies with a PMI of less than 8h to avoid potential loss of signal due to long PMI. Immunoblot and quantification analysis using both antibodies showed increased levels of Kidins220 in AD versus control samples ( Figure 1D and 1E), being differences more evident with Kidins220-Nt (p<0.01) compared to Kidins220-Ct (p<0.05). Furthermore, Kidins220-Nt revealed a positive correlation between Kidins220 and p-tau in AD samples (r 2 =0.236, n=17, p<0.05) ( Figure 1F). Finally, immunofluorescence of human AD brain showed partial co-localization of Kidins220 with p-tau ( Figure 1G).
Kidins220 is present in CSF samples from AD patients and correlates with tau. 9 In order to evaluate Kidins220 as a CSF biomarker, CSF samples were obtained from 12 controls, 5 patients with aMCI, 4 prodromal AD and 9 AD patients (see supplementary Table 1

for demographic details). Immunoblot analysis with
Kidins220-Nt showed Kidins220 presence, clearly more evident in some CSF samples from AD patients, while Kidins220-Ct rendered no specific bands ( Figure   2A). To confirm specificity of Kidins220 signal in CSF samples, we also tested a monoclonal antibody generated against 340 aminoacids of Kidins220 C-terminal region (Kidins220-M) [12]. This antibody detected FL-Kidins220 in those AD samples with higher Kidins220-Nt signal (Figure 2A). Although Kidins220-Nt labelled different bands in CSF it was difficult to undoubtedly identify the specific bands corresponding to the N-terminal fragments. Therefore, we only quantified Kidins220-FL-Nt band, and found it was significantly increased in CSF samples from AD patients ( Figure 2B). In addition, in these patients there was a positive correlation between CSF Kidins220-Nt

DISCUSSION
To evaluate Kidins220 as a potential biomarker for neurodegeneration, we have performed a quantitative study of Kidins220 in brain necropsies and CSF samples.
Kidins220 was detected in human brain by immunoblot using two polyclonal antibodies that recognise either the carboxy-or amino-termini of Kidins220. We found a negative correlation between Kidins220 levels and PMI in human brain necropsies, which was specific to the use of the C-terminal antibody. Moreover, only the N-terminal Kidins220 signal correlated with p-tau levels in AD brains. Importantly, Kidins220-Nt antibody detected Kidins220 in CSF samples from AD patients where it correlated positively with p-tau and tau content. Our results suggest that Kidins220 could constitute a novel marker of AD neurodegeneration.
In some CSF samples Kidins220 band was also visible using a monoclonal antibody raised against a big portion of Kidins220 C-terminal region but not with Kidins220-Ct that recognizes only last 17 aminoacids. These findings indicate that CSF may contain mainly N-terminal fragments of Kidins220 lacking the very C-terminal end after cleavage at the major identified calpain site (see scheme in Figure 1A), as has been reported in the ischemic brain [18]. The presence of transmembrane proteins in CSF is possible since cell-derived small vesicles known as exosomes have been detected in this biological fluid [20]. Indeed, transmembrane amyloid-related protein, as well as cytosolic proteins associated to exosomes such as tau, are found in CSF [21,22]. In addition, excitotoxic calpain-derived Kidins220 N-terminal fragments could be formed intracellularly during neurodegeneration and released to the extracellular space, associated to exosomes or free, as a consequence of neuronal death.
Our data demonstrate that Kidins220-Nt is a better tool for obtaining accurate and